Efficient Structure-Aware Image Smoothingby Local Extrema on Space-Filling Curve

Area of Science:

• Computer Vision
• Image Processing
• Signal Processing

Background:

• Traditional image smoothing methods often struggle with preserving structural details and can introduce artifacts.
• Existing techniques may lack efficiency due to processing on a 2D image grid.

Purpose of the Study:

• To propose a novel, efficient, and structure-aware image smoothing approach.
• To reduce edge staircase-like artifacts commonly seen in image smoothing.
• To enhance the preservation of salient edges and details during smoothing.

Main Methods:

• Utilizing a space-filling curve as a reduced domain for processing image data.
• Applying empirical mode decomposition (EMD) and modulating local extrema.
• Implementing a joint contrast-based filter to consolidate edge structures.
• Employing explicit interpolants on the curve for fast envelope computation.

Main Results:

• The proposed method effectively separates edges and details, achieving a structure-aware smoothing effect.
• Dimensional reduction improves overall structure awareness by exploiting more extrema.
• The approach significantly reduces edge staircase-like artifacts.
• Experimental results show superior performance compared to state-of-the-art methods.

Conclusions:

• The novel image smoothing approach using space-filling curves is effective and efficient.
• The method enhances structure awareness and minimizes artifacts, offering practical advantages.
• This technique represents a significant advancement in image processing capabilities.